Intelligent Operational Optimization Method for Salt Lake Chemical Industrial Processes Based on Data-Driven Modeling

The process control performance of the washing process heavily influences the quality of products in the Salt Lake Chemical industrial process, which is mainly determined by two subtasks: process modeling and operation optimization. For these two tasks, this paper first demarcates the modeling and optimization objectives of the washing process by profoundly analyzing the process mechanism and operating rules. To fully consider the working condition type and sampling time, a novel global multi-condition-aware transformer (GMCA-Former) network is proposed for the process modeling task. After that, the relatively accurate GMCA-Former is used to simulate the process state. Next, a hybrid operational optimization strategy based on case matching and NSGA-II is applied to find the optimal operational parameter settings efficiently and stably. Notably, an event-triggered process model update mechanism is also designed to ensure the accuracy of the process model. Finally, the effectiveness of the proposed methods is verified in the actual industrial dataset.